Identity recognition method, device, terminal equipment and storage medium

Abstract

The present invention discloses an identity recognition method, which is applied to a monitoring system including at least two cameras. The method includes: obtaining an image to be recognized collected by a camera at a preset time; determining a person to be recognized from the image to be recognized, and obtaining a person who meets preset conditions from a preset person feature database according to the human features of the person to be recognized, so as to obtain a set of candidate persons; judging whether the person to be recognized matches the candidate person according to the preset image feature set; if so, taking the candidate person as the final recognition result of the person to be recognized. In the present invention, the set of candidate persons is first determined according to the human features of the person to be recognized, and then the final recognition result is determined from the set of candidate persons according to the preset image feature set, so as to improve the accuracy of identity recognition. The present invention also discloses an identity recognition device, a terminal device and a storage medium, which improve the accuracy of identity recognition.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to an identity recognition method, device, terminal equipment and storage medium. Background Art

[0002] As the country's security requirements continue to increase, identity recognition technology is becoming more and more popular. For example, for monitoring systems, it is necessary to identify the identity of people appearing in monitoring videos. Traditional identity recognition solutions based on video analysis may have low accuracy rates due to scenes such as people accidentally lowering their heads, blocking each other during activities, uneven light distribution, different camera angles and image quality, and similar seasonal clothing.

[0003] Therefore, the existing identification methods have the problem of low identity recognition accuracy. Summary of the invention

[0004] The main purpose of the present invention is to provide an identity recognition method, device, terminal equipment and storage medium, aiming to solve the technical problem of low accuracy of existing identity recognition.

[0005] To achieve the above object, the present invention proposes an identity recognition method, which is applied to a monitoring system, wherein the monitoring system includes at least two cameras, and the identity recognition method includes the following steps:

[0006] Obtaining an image to be identified captured by the camera at a preset time;

[0007] Determining a human body to be identified from the image to be identified;

[0008] Acquiring the human features of the human body to be identified;

[0009] According to the human features of the human body to be identified, a person who meets the preset conditions is obtained from a preset human feature database to obtain a set of candidate persons; the preset human feature database includes the identity of the person and the corresponding human features;

[0010] According to a preset image feature set, determining whether the human body to be identified matches a candidate person in the candidate person set; the preset image feature set includes: a first historical image feature set collected by the same camera and a second historical image feature set collected by a different camera;

[0011] If so, the matched candidate person is used as the final recognition result of the human body to be recognized.

[0012] In addition, to achieve the above-mentioned purpose, the present invention further proposes an identity recognition device, which is applied to a monitoring system, wherein the monitoring system includes at least two cameras, and the identity recognition device includes:

[0013] A first acquisition module, used to acquire the image to be recognized collected by the camera at a preset time;

[0014] A first determination module is used to determine a human body to be identified from the image to be identified;

[0015] A second acquisition module is used to acquire the human features of the human body to be identified;

[0016] A third acquisition module is used to acquire a person who meets a preset condition from a preset human feature database according to the human features of the person to be identified, so as to obtain a set of candidate persons; the preset human feature database includes the identity of the person and the corresponding human features;

[0017] A judgment module, used to judge whether the human body to be identified matches a candidate person in the candidate person set according to a preset image feature set; the preset image feature set includes: a first historical image feature set collected by the same camera and a second historical image feature set collected by different cameras;

[0018] The second determination module is used to, if yes, take the matched candidate person as the final recognition result of the human body to be recognized.

[0019] In addition, to achieve the above-mentioned purpose, the present invention also proposes a terminal device, characterized in that the terminal device includes: a memory, a processor, and an identity recognition program stored in the memory and running on the processor, and when the identity recognition program is executed by the processor, the steps of the identity recognition method described in any one of the above items are implemented.

[0020] In addition, to achieve the above-mentioned purpose, the present invention also proposes a storage medium, characterized in that an identity recognition program is stored on the storage medium, and when the identity recognition program is executed by a processor, the steps of any of the above-mentioned identity recognition methods are implemented.

[0021] The technical solution of the present invention adopts an identity recognition method, which is applied to a monitoring system. The monitoring system includes at least two cameras, and the identity recognition method includes the following steps: obtaining an image to be recognized collected by a camera at a preset time; determining a human body to be recognized from the image to be recognized; obtaining human body features of the human body to be recognized; according to the human body features of the human body to be recognized, obtaining a person who meets preset conditions from a preset human body feature database to obtain a candidate person set; the preset human body feature database includes the identity of the person and the corresponding human body features; according to the preset image feature set, judging whether the human body to be recognized matches the candidate person in the candidate person set; the preset image feature set includes: a first historical image feature set collected by the same camera and a second historical image feature set collected by different cameras; if so, taking the matched candidate person as the final recognition result of the human body to be recognized; that is, in the present invention, firstly, according to the human body features of the human body to be recognized, the candidate person set is determined, and then according to the first historical image feature set collected by the same camera and the second historical image feature set collected by different cameras, the final recognition result is determined from the candidate person set, that is, according to the historical image feature set, the person recognized based on the human body features is further verified, and compared with the solution of only recognizing based on the face, the accuracy of identity recognition can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0023] Figure 1 A schematic diagram of the structure of a terminal device in a hardware operating environment involved in an embodiment of the present invention;

[0024] Figure 2 A schematic diagram of a flow chart of an identity recognition method according to a first embodiment of the present invention;

[0025] Figure 3 A schematic diagram of determining a human body image to be identified according to the first embodiment of the present invention;

[0026] Figure 4 A schematic diagram of a process of obtaining human body features according to a first embodiment of the present invention;

[0027] Figure 5 A schematic diagram of a process for obtaining geographic location information of a target person according to a ninth embodiment of the present invention;

[0028] Figure 6 A schematic diagram of determining an image to be recognized according to a ninth embodiment of the present invention;

[0029] Figure 7 It is a module schematic diagram of the identity recognition device of the present invention.

[0030] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0031] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0032] Reference Figure 1 , Figure 1 A schematic diagram of the terminal device structure of the hardware operating environment involved in the embodiment of the present invention.

[0033] The terminal device can be a user equipment (UE) such as a mobile phone, a smart phone, a laptop, a digital broadcast receiver, a personal digital assistant (PDA), a tablet computer (PAD), a handheld device, a vehicle-mounted device, a wearable device, a computing device, a smart home device (such as a smart TV, a refrigerator, a washing machine, an air conditioner, a range hood, etc.), a monitoring device or other processing device connected to a wireless modem, a mobile station (MS), etc. The terminal device station can be a control panel light smart home device. The device may be called a user terminal, a portable terminal, a desktop terminal, etc.

[0034] Typically, the terminal device includes: at least one processor 301, a memory 302, and an identity recognition program stored in the memory and executable on the processor, wherein the identity recognition program is configured to implement the steps of the identity recognition method described in any of the following embodiments.

[0035] The processor 301 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 301 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 301 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 301 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. The processor 301 may also include an AI (Artificial Intelligence) processor, which is used to process operations related to the identity recognition method, so that the identity recognition method model can be trained and learned autonomously to improve efficiency and accuracy.

[0036] The memory 302 may include one or more computer-readable storage media, which may be non-transitory. The memory 302 may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 302 is used to store at least one instruction, which is used to be executed by the processor 301 to implement the identity recognition method provided in the method embodiment of the present application.

[0037] In some embodiments, the terminal may further optionally include: a communication interface 303 and at least one peripheral device. The processor 301, the memory 302 and the communication interface 303 may be connected via a bus or a signal line. Each peripheral device may be connected to the communication interface 303 via a bus, a signal line or a circuit board. Specifically, the peripheral device includes: at least one of a radio frequency circuit 304, a display screen 305 and a power supply 306.

[0038] The communication interface 303 may be used to connect at least one peripheral device related to I / O (Input / Output) to the processor 301 and the memory 302. In some embodiments, the processor 301, the memory 302, and the communication interface 303 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 301, the memory 302, and the communication interface 303 may be implemented on a separate chip or circuit board, which is not limited in this embodiment.

[0039] The radio frequency circuit 304 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 304 communicates with the communication network and other communication devices through electromagnetic signals. The radio frequency circuit 304 converts the electrical signal into an electromagnetic signal for transmission, or converts the received electromagnetic signal into an electrical signal. Optionally, the radio frequency circuit 304 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a user identity module card, and the like. The radio frequency circuit 304 can communicate with other terminals through at least one wireless communication protocol. The wireless communication protocol includes, but is not limited to: a metropolitan area network, various generations of mobile communication networks (2G, 3G, 4G and 5G), a wireless local area network and / or a WIFI (Wireless Fidelity) network. In some embodiments, the radio frequency circuit 304 may also include circuits related to NFC (Near Field Communication), which is not limited in this application.

[0040] The display screen 305 is used to display the UI (User Interface). The UI may include graphics, text, icons, videos and any combination thereof. When the display screen 305 is a touch display screen, the display screen 305 also has the ability to collect touch signals on the surface or above the surface of the display screen 305. The touch signal can be input to the processor 301 as a control signal for processing. At this time, the display screen 305 can also be used to provide virtual buttons and / or virtual keyboards, also known as soft buttons and / or soft keyboards. In some embodiments, the display screen 305 can be one, the front panel of the electronic device; in other embodiments, the display screen 305 can be at least two, respectively arranged on different surfaces of the electronic device or in a folding design; in some further embodiments, the display screen 305 can be a flexible display screen, arranged on a curved surface or a folding surface of the electronic device. Even, the display screen 305 can also be set to a non-rectangular irregular shape, that is, a special-shaped screen. The display screen 305 can be made of materials such as LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode, organic light-emitting diode).

[0041] The power supply 306 is used to power various components in the electronic device. The power supply 306 can be AC ​​power, DC power, a disposable battery, or a rechargeable battery. When the power supply 306 includes a rechargeable battery, the rechargeable battery can support wired charging or wireless charging. The rechargeable battery can also be used to support fast charging technology. Those skilled in the art will understand that Figure 1 The structure shown in the figure does not constitute a limitation on the terminal device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0042] In addition, an embodiment of the present invention further proposes a storage medium, on which an identity recognition program is stored, and when the identity recognition program is executed by a processor, the steps of the identity recognition method described in any of the following embodiments are implemented. Therefore, it will not be repeated here. In addition, the description of the beneficial effects of adopting the same method will not be repeated. For technical details not disclosed in the computer-readable storage medium embodiment involved in this application, please refer to the description of the method embodiment of this application. As an example, the program instructions can be deployed to be executed on one computing device, or on multiple computing devices located at one location, or on multiple computing devices distributed at multiple locations and interconnected by a communication network.

[0043] A person skilled in the art can understand that the implementation of all or part of the process in any of the following embodiments can be completed by instructing the relevant hardware through a computer program, and the first identity recognition program or the second identity recognition program can be stored in a computer-readable storage medium, and when the program is executed, it can include the process of the embodiments of the following methods. The above-mentioned storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0044] Based on the aforementioned hardware structure, an embodiment of the identity recognition method of the present invention is proposed.

[0045] Reference Figure 2 , Figure 2 This is a flow chart of a first embodiment of an identity recognition method of the present invention. The identity recognition method comprises the following steps:

[0046] Step S11: Acquire the image to be recognized captured by the camera at a preset time.

[0047] It should be noted that the identity recognition method provided in this embodiment is applied to a monitoring system, which includes at least two cameras, and the cameras can be installed in the corresponding area according to actual needs to monitor the area. The camera can be installed indoors to monitor the indoor environment; the camera can also be installed outdoors to monitor the outdoor environment.

[0048] The preset time can be a pre-set time, which can be a specific time; it can also be a time determined according to a preset acquisition frequency. The preset acquisition frequency can be flexibly set according to actual needs. For example, assuming that the acquisition frequency is 30 seconds / time, that is, the image to be recognized captured by the camera is acquired every 30 seconds. Assuming that the last time the image to be recognized was acquired by the camera was 12:00:00, the next time the image to be recognized is acquired, the image to be recognized is acquired by the camera at 12:00:30.

[0049] The camera in the monitoring system will collect images during the monitoring process. In this embodiment, the image collected by the camera in the monitoring system at a preset time is obtained as the image to be identified. For the sake of distinction, in this embodiment, the camera that collects the image to be identified is recorded as the first camera, and the camera other than the first camera is recorded as the second camera.

[0050] Step S12: Determine a human body to be identified from the image to be identified.

[0051] In this embodiment, human body detection is performed on the image to be identified, so as to determine the human body to be identified from the image to be identified.

[0052] In some implementations, step S12 includes: determining a human body image to be identified of a human body to be identified from the images to be identified.

[0053] After performing human body detection on the image to be identified and determining the human body to be identified, the area where the human body to be identified is located is captured as the human body image to be identified. The specific extraction method can be flexibly set according to actual needs.

[0054] In some embodiments, step S12 includes:

[0055] Step 1: Perform human body detection on the image to be identified to obtain the number of human body detection frames, the coordinates of the human body detection frames, and the coordinates of the human body mask area.

[0056] Among them, the human body detection frame is a rectangular frame, the human body detection frame includes the human body, one human body detection frame includes one human body, the number of human body detection frames is the number of human bodies, the coordinates of the human body detection frames are the coordinates corresponding to the area where the human body is located; the human body mask area is the area outside the human body contour.

[0057] Step 2: According to the number of human detection frames, the coordinates of the human detection frames, and the coordinates of the human mask area, the area within the human detection frame is captured as the human image to be identified.

[0058] It should be noted that the area inside the human body detection frame and outside the human body contour is masked by the human body mask, so after interception, this part can be filled with zero pixels. Figure 3 As shown, assuming that in the image to be identified, there is a human body 102 in the human body detection frame 101, and the area outside the human body 102 is masked by the human body mask (i.e., the area filled with diagonal lines in the figure), after intercepting the area of ​​the human body detection frame 101, the area outside the human body 102 and inside the human body detection frame 101 is filled with zero pixels.

[0059] Step S13: Acquire the body features of the person to be identified.

[0060] In this embodiment, after determining the human body to be identified, the human body features of the human body to be identified are obtained. The categories of human body features include but are not limited to facial features, human body attribute features, and pedestrian features. It should be noted that facial features are facial features, and human body attribute features include but are not limited to the type and color of clothes.

[0061] In some embodiments, in order to improve the accuracy of recognition, step S13 includes: obtaining at least two human features of the person to be recognized. For example, facial features, human attribute features, and pedestrian features of the person to be recognized can be obtained. In this way, in subsequent recognition, compared with the existing situation of performing identity recognition only by facial features, performing identity recognition by at least two human features can avoid the problem of being unable to recognize a person in the case of an incomplete face (for example, wearing a mask, a hat, a bad angle, etc., resulting in an incomplete face in the collected image), changing clothes, etc., thereby improving the accuracy of identity recognition.

[0062] In this embodiment, the method of acquiring the human features of the person to be identified can be flexibly set according to actual needs.

[0063] For example, in some implementations, the human features of the person to be identified may be directly acquired from the image to be identified to obtain the human features to be identified.

[0064] For another example, in some embodiments, when step S12 includes determining a human body image of a human body to be identified from the image to be identified, step S13 includes: obtaining at least two human body features of the human body to be identified from the human body image to be identified. For example, in one example, step S13 includes: extracting facial features, human body attribute features, and pedestrian features from the human body image to be identified.

[0065] When extracting facial features, human attribute features and pedestrian features from a human image to be identified, considering that the human image to be identified may have incomplete faces or incomplete human bodies, human features cannot be extracted, or the extracted human features are unusable, wasting power. In some embodiments, see Figure 4 As shown, step S13 includes:

[0066] Step S131: performing face detection on the human image to be identified to obtain a face integrity value.

[0067] Among them, the face integrity value is used to evaluate the integrity of the face.

[0068] Step S132: Determine whether the face integrity value is greater than a preset face integrity threshold.

[0069] The face completeness threshold can be flexibly set according to actual needs.

[0070] Step S133: If yes, extract facial features from the image of the person to be identified.

[0071] Since the face integrity of the face in the human image to be identified is greater than the preset face integrity threshold, it indicates that the face in the human image to be identified is relatively complete, and the face features of the face in the human image to be identified are extracted. The specific method of extracting face features can be flexibly set according to actual needs.

[0072] For example, in some embodiments, in step S131, when performing face detection on the human image to be identified, the coordinates of the face detection frame can be obtained. Step S133 may include: according to the coordinates of the face detection frame, intercepting the part within the face detection frame from the human image to be identified as the face image to be identified; performing 3D face correction on the face image to be identified, and normalizing its image size, and inputting the normalized face image to be identified into the face feature extraction model to obtain face features. Among them, if the face detection frame exceeds the range of the human image to be identified, it can be padded with zero pixels. Among them, the face feature extraction model is a pre-trained neural network model for extracting face features.

[0073] Step S134: determining whether the human body integrity value of the human body in the human body image to be identified is greater than a preset human body integrity threshold.

[0074] The human body integrity value is used to characterize the human body integrity; the preset human body integrity threshold can be flexibly set according to actual needs.

[0075] In this embodiment, the human body integrity value of the human body in the human body image to be identified is obtained, and is compared with a preset human body integrity threshold to determine whether the human body integrity value is greater than the preset human body integrity threshold.

[0076] Step S135: If yes, extract pedestrian features and human attribute features from the human image to be identified.

[0077] Since the human body integrity of the human body in the human body image to be identified is greater than the preset human body integrity threshold, it indicates that the human body in the human body image to be identified is relatively complete, and the human body attribute features and pedestrian features of the human body in the human body image to be identified are extracted. The specific method of extracting the human body attribute features and pedestrian features can be flexibly set according to actual needs.

[0078] For example, in some embodiments, the method of extracting pedestrian features and human attribute features from the human image to be identified includes: normalizing the size of the human image to be identified, inputting the normalized human image to be identified into a human attribute feature extraction model to obtain human attribute features; inputting the normalized human image to be identified into a re-id (Person Re-identification) model to obtain pedestrian features. The human attribute feature extraction model is a pre-trained neural network model for extracting human attribute features; and the re-id model is a pre-trained neural network model for analyzing pedestrian features.

[0079] In this embodiment, the execution order of step S131 and step 134 is not limited. In some embodiments, step S131 and step 134 can be executed simultaneously; in some embodiments, step S131 can be executed first, and then step 134; in some embodiments, step S134 can be executed first, and then step 131.

[0080] Step S14: According to the human features of the person to be identified, persons meeting preset conditions are obtained from a preset human feature database to obtain a set of candidate persons.

[0081] The preset portrait feature database includes the identity of the person and the body features corresponding to the person.

[0082] Among them, the preset conditions can be flexibly set according to actual needs.

[0083] Step S15: judging whether the human body to be identified matches a candidate person in the candidate person set according to a preset image feature set.

[0084] The preset image feature set includes: a first historical image feature set collected by the same camera and a second historical image feature set collected by a different camera. The same camera is a camera that collects the image to be identified, and the different camera is a camera different from the camera that collects the image to be identified. It should be understood that the same camera and the different cameras are both cameras in the monitoring system.

[0085] In some embodiments, the first historical image feature set may include: a first image captured by the first camera at a preset time and a first storage time period before the preset time, a first human feature extracted from the first image, and a person corresponding to the first human feature. The first storage time period may be flexibly set according to actual needs, for example, it may be set to 3 minutes, 2 minutes, etc. In one example, assuming that the preset time is 12:00:00 and the first storage time period is 3 minutes, the first historical image feature set includes: a first image captured by the first camera at 11:57:00-12:00:00, a first human feature extracted from the first image, and a person corresponding to the first human feature.

[0086] In some embodiments, the second historical image feature set may include: images captured by the second camera at a preset time and a second storage time period before the preset time, human features extracted from the images, and persons corresponding to the human features. The second storage time period may be flexibly set according to actual needs, for example, it may be set to 4 minutes, 5 minutes, etc. The first storage time period and the second storage time period may be consistent or inconsistent (that is, the storage time periods corresponding to the cameras are inconsistent).

[0087] Regardless of whether it is a person corresponding to the first human feature or a person corresponding to the second human feature, it can be a candidate person obtained from a preset human feature database based on the human feature, or it can be a person matched with the human feature determined based on a preset image feature set.

[0088] Since the preset image feature set includes: the first image captured by the first camera at a preset time (i.e., the image to be identified), the human features extracted from the image, and the persons corresponding to the human features, after step S14 and before step S15, the image to be identified, the features corresponding to the human body to be identified extracted from the image to be identified, and the candidate persons for the human body to be identified can also be updated to the preset image feature set.

[0089] In some embodiments, the image feature set may also include the acquisition time of each human feature, the recognition status, etc., wherein the recognition status includes preliminary recognition and final recognition, wherein preliminary recognition is a state in which the final recognition result has not been obtained, and final recognition is a state in which the final recognition result has been obtained, that is, the state in which step S16 has been executed.

[0090] For example, see Table 1, which is an example of an image feature set.

[0091] Table 1

[0092]

[0093] In some embodiments, step S15 includes but is not limited to the following:

[0094] The first one: Since the same camera usually captures the same person multiple times within a certain time range, it is possible to determine whether there is a candidate person among the persons corresponding to the previous frame image based on a preset image feature set, wherein the previous frame image is the first image captured by the first camera before a preset moment; if so, it indicates that the same camera captured the candidate person at different moments, which largely indicates that the person to be identified in the image to be identified is the candidate person, and it is determined that the candidate person matches the human body to be identified; if not, it indicates that the same camera has not captured the candidate person within a period of time, which largely indicates that the person to be identified in the image to be identified is not the candidate person, and it is determined that the candidate person does not match the human body to be identified; that is, by determining whether there is a candidate person in the first image captured by the same camera before the preset moment, it is determined whether the candidate person matches the human body to be identified. For example, referring to Table 1, assuming that the preset time is 11:55:40, the first camera is camera 1, the image to be identified is image 2, the human features of the human body to be identified are human features 2, and the candidate persons are person 2 and person 3. Then, it is determined whether person 2 and person 3 exist among the persons corresponding to image 1 (image 1 is the previous frame image). Since person 2 exists among the persons corresponding to image 1, person 2 matches the human body to be identified; since person 3 does not exist among the persons corresponding to image 1, person 3 does not match the human body to be identified.

[0095] The second type: Since cameras with the same coverage will capture the same person at the same time, it is possible to determine whether there is a candidate person among the people corresponding to the first associated image based on a preset image feature set, wherein the first associated image is an image captured at a preset time by a second camera with the same coverage area as the first camera; if there is a candidate person among the people corresponding to the first associated image, it indicates that cameras with the same coverage area all captured the candidate person at the same time, which largely indicates that the person to be identified on the image to be identified is the candidate person, and it is determined that the candidate person matches the person to be identified; if there is no candidate person among the people corresponding to the first associated image, it indicates that other cameras with the same coverage area did not capture the candidate person at the same time, which largely indicates that the person to be identified on the image to be identified is not the candidate person, and it is determined that the candidate person does not match the person to be identified. In other words, by determining whether there is a candidate person in the second image captured at a preset time by a camera with the same coverage area, it is determined whether the candidate person matches the person to be identified. For example, referring to Table 1, assuming that the preset time is 11:55:40, the first camera is camera 1, the image to be identified is image 2, the human body to be identified is the human body corresponding to human feature 2, the candidate persons are person 2 and person 3, and the coverage range of camera 2 is the same as that of camera 1. Then, the first associated image is image 4. Since person 2 exists in image 4, person 2 matches the human body to be identified; person 3 does not exist in image 4, therefore, person 3 does not match the human body to be identified.

[0096] The third type: Since cameras with different coverage ranges capture different persons at the same time, it is possible to determine whether there is a candidate person among the persons corresponding to the second associated image according to a preset image feature set. The second associated image is an image captured at a preset time by a camera with a different coverage area from the first camera; if there is no candidate person among the persons corresponding to the second associated image, it is determined that the candidate person matches the human body to be identified; if there is a candidate person among the persons corresponding to the second associated image, since cameras with different coverage ranges cannot capture the same person at the same time, it is determined that the candidate person does not match the human body to be identified. In other words, by determining whether there is no candidate person in the second image captured at a preset time by cameras with different coverage ranges, it is determined whether the candidate person matches the human body to be identified. For example, as shown in Table 1, assuming that the preset time is 11:55:40, the first camera is camera 1, the image to be identified is image 2, the human body to be identified is the human body corresponding to human body feature 2, the candidate persons are persons 2 and 3, the coverage of camera 3 is different from that of camera 1, and the second associated image is image 6. Since persons 3 and 4 exist in image 6, person 3 does not match the human body to be identified, and person 2 matches the human body to be identified.

[0097] The fourth type: Since the activity trajectory of a person must move from one place to an adjacent place, it is possible to determine whether the coverage area of ​​the target camera and the first camera in the image feature set is adjacent based on the preset image feature set, wherein the target camera is the camera of the candidate person in the second image captured outside the preset time; if adjacent, then match, if not adjacent, then do not match. Among them, if there is only one target camera, then determine whether the coverage area of ​​the target camera and the first camera is adjacent, and if there are at least two target cameras, then determine whether any of the target camera and the first camera has a camera with an adjacent coverage area.

[0098] In some embodiments, in order to further improve the accuracy of the judgment, if there are at least two target cameras, it is determined whether the coverage areas of the two cameras with adjacent first acquisition times are adjacent in the target camera and the first camera, wherein the first acquisition time here is the acquisition time of the image of the candidate person in the target camera and the first camera. For example, as shown in Table 1, assuming that the preset time is 11:55:40, the first camera is camera 1, the image to be identified is image 2, the human body to be identified is the human body corresponding to human body feature 2, and the candidate persons are person 2 and person 3. Since person 2 exists in the images corresponding to camera 4 and camera 5, the target cameras are cameras 4 and 5, the acquisition time of image 7 is adjacent to that of image 9, and the acquisition time of image 9 is adjacent to that of image 2. It is determined whether the coverage areas of camera 4 and camera 5 are adjacent, and it is determined whether the coverage areas of camera 5 and camera 1 are adjacent. If the coverage areas of camera 4 and camera 5 are adjacent, and the coverage areas of camera 5 and camera 1 are adjacent, then person 2 matches camera 1.

[0099] It should be noted that the above four methods can also be combined. If any two of the above four methods are combined, when the matching result of one of the methods is that the candidate person matches the human body to be identified, the candidate person can be determined to match the human body to be identified, or, when the matching results of at least two methods are that the candidate person matches the human body to be identified, the candidate person can be determined to match the human body to be identified. For example, assuming that it is a combination of the first method and the second method, for candidate person 1, when it is determined that candidate person 1 exists in the personnel corresponding to the previous frame image or candidate person 1 exists in the personnel corresponding to the first associated image, it can be determined that candidate person 1 matches the human body image to be identified; or, when it is determined that candidate person 1 exists in the personnel corresponding to the previous frame image and candidate person 1 exists in the personnel corresponding to the first associated image, it can be determined that candidate person 1 matches the human body image to be identified.

[0100] Step S16: If yes, the matched candidate person is used as the final recognition result of the human body to be recognized.

[0101] If the person to be identified matches the candidate person, the candidate person that matches the person to be identified is used as the final recognition result of the person to be identified.

[0102] The identity recognition method provided in this embodiment is applied to a monitoring system, and the monitoring system includes at least two cameras. The identity recognition method includes the following steps: obtaining an image to be recognized captured by a camera at a preset time; determining a person to be recognized from the image to be recognized; obtaining the human features of the person to be recognized; according to the human features of the person to be recognized, obtaining a person who meets the preset conditions from a preset human feature database to obtain a set of candidate persons; the preset human feature database includes the identity of the person and the corresponding human features; according to the preset image feature set, judging whether the person to be recognized matches the candidate person in the candidate person set; the preset image feature set includes: a first historical image feature set captured by the same camera and a second historical image feature set captured by different cameras; if so, taking the matching candidate person as the final recognition result of the person to be recognized; that is, in this embodiment, firstly, according to the human features of the person to be recognized, a set of candidate persons is determined, and then according to the first historical image feature set captured by the same camera and the second historical image feature set captured by different cameras, a final recognition result is determined from the set of candidate persons, that is, according to the historical image feature set, the person recognized based on the human features is further verified, and compared with the scheme of recognizing only based on facial features, the accuracy of identity recognition can be improved.

[0103] Based on the first embodiment, a second embodiment of the identity recognition method of the present invention is proposed. In this embodiment, after step S11 and before step S12, the identity recognition method further includes the following steps:

[0104] Step S17: Evaluate the image quality of the image to be recognized to obtain an image quality evaluation value.

[0105] The image quality evaluation value is used to characterize the quality of an image. The larger the image quality evaluation value is, the better the image quality is.

[0106] In this embodiment, the image to be identified can be input into an image quality assessment model to obtain an image quality assessment value, wherein the image quality assessment model is a pre-trained neural network model for assessing image quality.

[0107] Step S18: Determine whether the image quality evaluation value is greater than a preset image quality threshold.

[0108] The preset image quality threshold is a preset image quality value. The specific value of the preset image quality threshold can be flexibly set according to actual needs.

[0109] Step S19: If not, pre-process the image to be recognized.

[0110] If the image quality evaluation value of the image to be identified is less than the preset image quality threshold, it indicates that the image quality of the image to be identified is not good, for example, the image may be blurry, the noise is large, etc. Therefore, in order to avoid subsequent unrecognition, the image to be identified is preprocessed to improve the image quality. Among them, the preprocessing includes but is not limited to at least one of the following methods: deblurring the image to be identified; texture enhancement processing, denoising processing, light balance processing, etc. Step S19 includes at least one of the following methods: deblurring the image to be identified; texture enhancement processing of the image to be identified; denoising processing of the image to be identified; light balance processing of the image to be identified. When the preprocessing method includes at least two processing methods, the execution order of each processing method can be flexibly set according to actual needs. For example, in one example, the preprocessing includes: deblurring processing, texture enhancement processing, denoising processing, and light balance processing, and the preprocessing of the image to be identified includes: first deblurring the image to be identified; and then texture enhancement processing, denoising processing, and light balance processing of the image to be identified after deblurring.

[0111] In this embodiment, step S12 includes: determining the human body to be identified from the preprocessed image to be identified. Since the image to be identified is preprocessed before step S12, in step S12, the human body to be identified is determined from the preprocessed image to be identified.

[0112] In other embodiments, the judgment result of step S18 is: the image quality evaluation value of the image to be identified is greater than the preset image quality threshold. This indicates that the image quality of the image to be identified is good, and therefore, it is possible to directly proceed to S12, that is, after step S18, the identity recognition method includes: If yes, directly proceed to step S12.

[0113] The identity recognition method provided in the present embodiment, when acquiring an image to be recognized, first evaluates the image quality of the image to be recognized to obtain an image quality evaluation value, and when the image quality evaluation value is less than or equal to a preset image quality threshold, preprocesses the image to be recognized, and based on the preprocessed image to be recognized, determines the human body to be recognized from the image to be recognized and subsequent steps. That is to say, in the present embodiment, if the quality of the acquired image to be recognized is not good, the image to be recognized will be preprocessed, thereby improving the image quality of the image to be recognized, improving the accuracy of identity recognition, and avoiding the situation where the image to be recognized captured by the camera has poor quality and cannot be recognized due to reasons such as uneven light distribution.

[0114] Based on the first embodiment, a third embodiment of the identity recognition method of the present invention is proposed. In this embodiment, step S13 includes: obtaining facial features, human attribute features, and pedestrian features of a person to be recognized.

[0115] In this embodiment, step S14 includes: searching from a preset human feature database for persons whose facial feature similarity with the person to be identified is greater than a preset first facial feature similarity threshold, whose human attribute feature similarity is greater than a preset first human attribute feature similarity threshold, and whose pedestrian feature similarity is greater than a preset first pedestrian feature similarity threshold, so as to obtain a set of candidate persons.

[0116] That is to say, the facial features of the person to be identified are compared with the facial features in a preset human feature database, the human attribute features of the person to be identified are compared with the human attribute features in the preset human feature database, and the pedestrian features of the person to be identified are compared with the pedestrian features in the preset human feature database. Then, a person whose facial feature similarity with the person to be identified is greater than a preset first facial feature similarity threshold, whose human attribute feature similarity is greater than a preset first human attribute feature similarity threshold, and whose pedestrian feature similarity is greater than a preset first pedestrian feature similarity threshold is found as a candidate person in the candidate person set, or a person whose facial feature similarity with the person to be identified is greater than a preset first facial feature similarity threshold, whose human attribute feature similarity is greater than a preset first human attribute feature similarity threshold, or whose pedestrian feature similarity is greater than a preset first pedestrian feature similarity threshold is found as a candidate person in the candidate person set.

[0117] Among them, the preset first facial feature similarity threshold, the preset first human attribute feature similarity threshold, and the preset first pedestrian feature similarity threshold can be flexibly set according to actual needs.

[0118] Of course, in some implementations, if at least one of the facial features, body attribute features, and pedestrian features cannot be extracted (for example, the face is incomplete, etc.), the feature comparison will not be performed.

[0119] The identity recognition method provided in this embodiment searches for persons whose facial feature similarity with the person to be recognized is greater than a preset first facial feature similarity threshold, whose body attribute feature similarity is greater than a preset first body attribute feature similarity threshold, and whose pedestrian feature similarity is greater than a preset first pedestrian feature similarity threshold from a preset body feature database, so as to obtain a set of candidate persons. That is to say, the person to be recognized is recognized based on facial features, body attribute features, and pedestrian features, thereby improving the recognition accuracy and avoiding the situation where the identity cannot be recognized when the face is blocked due to recognition based only on facial features.

[0120] Based on the third embodiment, a fourth embodiment of the identity recognition method of the present invention is proposed. In this embodiment, before step S14, the identity recognition method further includes the following steps:

[0121] Step S20: collecting human body images.

[0122] Since in step S14, it is necessary to search for a person who meets the preset conditions from a preset human feature database, it is necessary to create a portrait feature database before step S14. In order to create the portrait feature database, in this embodiment, a human image is first collected. The human image can be an image including only a human face, or an image including a human face and a body.

[0123] Step S21: Acquire human body features from the human body image.

[0124] In this embodiment, after collecting the human body image, portrait features are extracted from the human body image, wherein the human body features include but are not limited to at least one of facial features, human body attribute features, pedestrian features, etc. In order to improve the accuracy of identity recognition, the human body features may include at least two portrait features. For example, the human body features may include facial features, human body attribute features and pedestrian features. At this time, step S20 includes: obtaining facial features, human body features and pedestrian features of the human body image as portrait features.

[0125] Step S22: Create a human feature database based on the human features and the persons corresponding to the human images.

[0126] That is to say, the human body feature database includes persons and the human body features corresponding to the persons.

[0127] The identity recognition method provided in this embodiment first collects a human body image, obtains human body features from the human body image, and creates a human body feature database based on the human body features and the persons corresponding to the human body images, so as to be used in the subsequent step of obtaining persons who meet preset conditions from the preset human body feature database based on the human body features of the person to be identified.

[0128] Based on the first embodiment, a fifth embodiment of the identity recognition method of the present invention is proposed. In this embodiment, before step S15, the identity recognition method further includes the following steps:

[0129] Step S23: Screening the candidate personnel in the candidate personnel set according to the preset personnel trajectory library.

[0130] The preset personnel trajectory library includes personnel and the trajectory of each personnel.

[0131] In this embodiment, after obtaining the set of candidate personnel, the candidate personnel in the set of candidate personnel are screened according to a preset personnel trajectory library.

[0132] In some embodiments, step S23 includes:

[0133] Step 1: According to the preset personnel trajectory library, the estimated geographic location of each candidate in the candidate set at the preset time is estimated.

[0134] According to the preset personnel trajectory, the geographical location that each candidate in the candidate set may appear at the preset time is estimated as the estimated geographical location.

[0135] For example, assuming that before the preset time, in the preset personnel trajectory library, the last geographical location of candidate 1 was room 1, and the rooms adjacent to room 1 were 2 and 3, then at the preset time, the possible geographical locations of candidate 1 are: rooms 1, 2, and 3; before the preset time, in the preset personnel trajectory library, the last geographical location of candidate 2 was room 4, and the rooms adjacent to room 4 were 5 and 6, then at the preset time, the possible geographical locations of candidate 2 are: rooms 4, 5, and 6.

[0136] Step 2: Delete the candidate whose estimated geographic location does not match the geographic location corresponding to the first camera.

[0137] It is determined whether the estimated geographic location of each candidate person includes the geographic location corresponding to the second camera; if so, the candidate person is retained; if not, the candidate person is deleted.

[0138] Continuing with the above example, assume that the geographic location corresponding to the first camera is Room 3. Therefore, delete candidate 2 and retain candidate 1.

[0139] Step S15 includes: judging whether the person to be identified matches the candidate person in the screened candidate person set according to the preset image feature set. Since the subsequent person is screened before step S15, in step S15, judging whether the person to be identified matches the candidate person in the screened candidate person set according to the preset image feature set.

[0140] In this embodiment, candidate persons in a candidate person set are screened according to a preset person trajectory library, and then subsequent recognition is performed according to the screened candidate person set to obtain a final recognition result, thereby improving recognition accuracy.

[0141] Based on the fifth embodiment, a sixth embodiment of the identity recognition method of the present invention is proposed. In this embodiment, after step S23 and before step S15, the identity recognition method further includes the following steps:

[0142] Step S25: Screening the set of candidate persons after screening according to the body features of the person to be identified.

[0143] In this embodiment, the candidate set screened according to the preset personnel trajectory library is further screened according to the human features of the person to be identified.

[0144] In some embodiments, step S25 includes:

[0145] Step 1: Determine whether the facial feature similarity value between the person to be identified and the first candidate is greater than a preset second facial feature similarity threshold.

[0146] Among them, the first candidate is the candidate with the highest similarity value with the facial features of the person to be identified in the candidate set screened according to the preset person trajectory library.

[0147] The second facial feature similarity threshold is greater than the first facial feature similarity threshold.

[0148] Step 2: If yes, only keep the first candidate.

[0149] If the facial feature similarity value between the person to be identified and the first candidate is greater than the preset second facial feature similarity threshold, the first candidate is retained and other candidate persons in the candidate set are deleted. In other words, the candidate set only includes the first candidate.

[0150] Step 3: If not, determine whether the pedestrian feature similarity value between the person to be identified and the second candidate person is greater than a preset second pedestrian feature similarity threshold.

[0151] The second candidate person is a candidate person with the highest similarity value with the pedestrian feature of the human body to be identified in the candidate person set screened according to the preset person trajectory library.

[0152] The preset second pedestrian feature similarity threshold is greater than the preset first pedestrian feature similarity threshold.

[0153] Step 4: If the pedestrian feature similarity value between the person to be identified and the second candidate person is greater than the preset second pedestrian feature similarity threshold, only the second candidate person is retained.

[0154] If the pedestrian feature similarity value between the person to be identified and the second candidate person is greater than the preset second pedestrian feature similarity threshold, the second candidate person is retained and other candidate persons in the candidate person set are deleted. In other words, the candidate person set only includes the second candidate person.

[0155] Step 5: If the pedestrian feature similarity value between the person to be identified and the second candidate person is less than or equal to the preset second pedestrian feature similarity threshold, determine whether the human attribute feature similarity value between the person to be identified and the third candidate person is greater than the preset second human attribute feature similarity threshold.

[0156] The third candidate person is a candidate person with the highest similarity value with the human attribute characteristics of the human body to be identified from the candidate person set screened according to the preset person trajectory library.

[0157] The preset second human attribute feature similarity threshold is greater than the preset first human attribute feature similarity threshold.

[0158] In some embodiments, before determining whether the similarity value of the human attribute feature between the person to be identified and the third candidate is greater than the preset second human attribute feature similarity threshold, it can also be determined whether there is a unique human attribute feature among the candidate persons, and the unique human attribute feature matches the human attribute feature of the person to be identified, wherein the unique human attribute feature is a human attribute feature that only one candidate person has and other candidate persons do not have; if so, the similarity value of the human attribute feature of the candidate person is increased, and the third candidate person is determined from the candidate persons according to the increased value of the human attribute feature similarity value. For example, assuming that the clothes of the person to be identified are white short sleeves, the similarity value of the human attribute feature of candidate person 1 is 10. Since among the candidate persons, only candidate person 1 is wearing white short sleeves, the similarity value of the human attribute feature of candidate person 1 is increased, and the third candidate person is determined from the candidate persons according to the increased similarity value of the human attribute feature.

[0159] Step 6: If the human attribute feature similarity value between the human to be identified and the third candidate is greater than a preset second human attribute feature similarity threshold, only the third candidate is retained.

[0160] That is to say, the candidate set only includes the third candidate.

[0161] Step S15 includes: judging whether the human body to be identified matches the candidate persons in the screened candidate person set according to a preset image feature set.

[0162] In this embodiment, the screened candidate person set is screened according to the human features of the human body to be identified, and then subsequent identification is performed based on the screened candidate person set to obtain a final identification result, thereby improving the accuracy of identification.

[0163] Based on the first embodiment, a seventh embodiment of the identity recognition method of the present invention is proposed. In this embodiment, after step S16, the identity recognition method further includes the following steps:

[0164] Step S25: updating the preset image feature set and the preset human feature database according to the human features of the human body to be identified and the final identification result.

[0165] In this embodiment, after the final recognition result is obtained, the preset human body feature database is updated according to the human body features of the human body to be recognized and the final recognition result.

[0166] In this embodiment, after obtaining the final recognition result, the preset image feature set is updated according to the human features of the human body to be recognized and the final recognition result. The updating of the preset image feature set includes but is not limited to at least one of the following methods:

[0167] The first method: updating the person and recognition status corresponding to the person to be identified in the preset image feature set. For example, referring to Table 1, assuming that the image to be identified is image 2, the human feature of the person to be identified is human feature 2, and the final recognition result is that the person to be identified is person 2, then the person corresponding to human feature 2 is updated to include only person 2, and the recognition status is updated to final recognition.

[0168] The second method: based on the human features of the person to be identified and the final recognition result, the person and recognition status corresponding to the human features that can determine the final recognition result are updated. For example, as shown in Table 1, the human features of the person to be identified are human features 2, and the final recognition result is that the person to be identified is person 2. Since the person corresponding to the image captured by camera 2 at the preset time is person 2, and the coverage area of ​​camera 1 and camera 2 is the same, it can be determined that the final recognition result of the person existing in the image captured by camera 2 at the preset time is person 2, and therefore, the recognition status corresponding to human feature 4 is updated to final recognition.

[0169] In this embodiment, after obtaining the final recognition result, the preset image feature set and the preset human feature database are updated according to the human features of the person to be identified and the final recognition result, so that the preset image feature set and the preset human feature database are kept up to date to improve the recognition accuracy.

[0170] Based on the first embodiment, an eighth embodiment of the identity recognition method of the present invention is proposed. In this embodiment, after step S16, the identity recognition method further includes the following steps:

[0171] Step S26: adjusting the corresponding parameters of each camera according to whether a final recognition result is obtained from the images collected by each camera within a preset time period.

[0172] The preset time period can be flexibly set according to actual needs, for example, it can be set to 10 minutes, 15 minutes, etc. The parameters corresponding to each camera include but are not limited to: image quality threshold, face integrity threshold, body integrity threshold, storage time period corresponding to each camera in the image feature set (first storage time period, second storage time period)

[0173] The adjustment methods may be:

[0174] The first method: If the camera does not obtain the final recognition result in the images collected within the preset time period, the image quality threshold, the face integrity threshold, the body integrity threshold are lowered, and the storage time period corresponding to the camera is extended to improve the accuracy of personnel recognition.

[0175] The second type: If the camera obtains the final recognition result in the image collected within the preset time period, the image quality threshold, the face integrity threshold, the body integrity threshold are increased, and the corresponding storage time period of the camera is shortened to maintain the recognition accuracy while reducing the data processing volume.

[0176] In this embodiment, after obtaining the final recognition result, the corresponding parameters of each camera are adjusted according to whether the final recognition result is obtained in the images collected by each camera within a preset time period, so as to improve the recognition accuracy and reduce the data processing amount.

[0177] Based on the first embodiment, a ninth embodiment of the identity recognition method of the present invention is proposed. In this embodiment, after step S16, the identity recognition method further includes the following steps:

[0178] Step S27: Obtain the geographic location information of the target person based on the image to be identified.

[0179] The target person is the final recognition result of the person to be recognized.

[0180] In this embodiment, different spatial information and geographical location information are preset, and the geographical location information of the target person is acquired according to the image to be identified.

[0181] In one embodiment, see Figure 5 As shown, step S27 includes:

[0182] Step S271: Acquire first spatial grid information of the first camera.

[0183] The first spatial grid information includes the spatial grids obtained by dividing the coverage area of ​​the first camera and the geographical locations corresponding to each spatial grid. That is to say, in this embodiment, the coverage area of ​​the first camera is pre-divided into multiple spatial grids, and corresponding geographical location information is set for each spatial grid. It should be understood that the coverage area of ​​the camera is the area that the camera can capture.

[0184] Step S272: determining a first geographical location according to the coordinates of the human body image to be identified in the image to be identified and the first spatial grid information.

[0185] In this embodiment, the corresponding spatial grid is determined according to the coordinates of the human body image to be identified in the image to be identified, and the first geographical location is determined according to the corresponding spatial grid and the first spatial grid information.

[0186] For example, see Figure 6 As shown, Figure 6In the figure, the coverage area of ​​the first camera is divided into four spatial grids. In the image 601 to be identified, the human body image to be identified is in the spatial grid 602. Therefore, the geographical location corresponding to the spatial grid 602 is obtained as the first geographical location.

[0187] Step S273: Determine whether the target person is identified in the reference image captured by the second camera.

[0188] In this embodiment, the reference image is an image captured by the second camera at a preset time. The recognition here is to obtain a final recognition result, that is, to determine whether the target person exists in the final recognition result of the reference image captured by the second camera.

[0189] The method for obtaining the final recognition result in the reference image may be the same as the method for obtaining the final recognition result in the image to be recognized, which will not be described in detail here.

[0190] Step S274: If yes, determine the second geographic location according to the coordinates of the human body corresponding to the target person in the reference image and the second spatial grid information corresponding to the second camera.

[0191] The second spatial grid information includes the spatial grids obtained by dividing the coverage area of ​​the second camera and the geographical locations corresponding to each spatial grid. That is to say, in this embodiment, the coverage area of ​​the second camera is divided into multiple spatial grids in advance, and the corresponding geographical location information is set for each spatial grid. It should be understood that the coverage area of ​​the camera is the area that the camera can capture.

[0192] The method for determining the second geographical location may refer to the method for determining the first geographical location, which will not be described in detail here.

[0193] If the target person is identified in the reference image captured by the second camera, the second geographic location is determined according to the coordinates of the human body corresponding to the target person in the reference image and the second spatial grid information corresponding to the second camera.

[0194] Step S275: Acquire the overlapping area of ​​the first geographical location and the second geographical location as the geographical location information of the target person.

[0195] In this embodiment, since the target person exists in the images captured by the first camera and the second camera, in order to improve accuracy, the overlapping area of ​​the first geographic location and the second geographic location is obtained as the geographic location information of the target person.

[0196] Step S276: If not, the pixel values ​​of the pixels corresponding to the human body to be identified in the image to be identified are set to zero to obtain a background image.

[0197] If the target person is not identified in the reference image captured by the second camera, the pixel values ​​of the pixels corresponding to the human body to be identified in the image to be identified are set to zero to obtain a background image.

[0198] Step S277: Determine whether an object is detected from the background image.

[0199] In this embodiment, the background image is detected by an object detection model to determine whether an object is detected from the background image. The object detection model is a pre-trained neural network model used to detect objects from an image.

[0200] Step S278: If not, use the first geographic location as the geographic location information of the target person.

[0201] If no object is detected from the background image, the first geographical location is directly used as the geographical location information of the target person.

[0202] Step S279: If yes, compare the object with a preset reference object.

[0203] If an object is detected from the background image, the detected object is compared with a preset reference object, wherein the preset reference object may be a pre-set reference object existing within the coverage range of the camera, for example, a potted plant, a table, or other object that is not easy to move.

[0204] Step S2710: If the object does not match the preset reference object, the first geographic location is used as the geographic location information of the target person.

[0205] If the object does not match the preset reference object, the first geographic location is directly used as the geographic location information of the target person.

[0206] Step S2711: If the target object matches the preset reference object, a third geographical location is determined according to the coordinates of the object in the image to be identified, the coordinates of the human body to be identified in the image to be identified, and the key geographical location corresponding to the preset reference object.

[0207] In this embodiment, each preset reference object is set with a corresponding geographical location, ie, a key geographical location.

[0208] Based on the coordinates of the object identified from the background image in the background image and the coordinates of the human body image to be identified, the difference between the two is determined, and the third geographical location is determined based on the difference and the key geographical location of a preset reference object corresponding to the object.

[0209] Step S2712: Obtain the overlapping area of ​​the first geographical location and the third geographical location as the geographical location information of the target person.

[0210] An overlapping area of ​​the first geographical location and the third geographical location is obtained as the location information of the target person.

[0211] Step S28: Update the preset personnel trajectory library according to the geographic location information of the target person.

[0212] After obtaining the location information of the target person, the preset person trajectory library is updated according to the geographic location information of the target person.

[0213] In this embodiment, a GIS (Geographic Information System) service module may be pre-set to store the aforementioned information related to the geographic location, such as the geographic location corresponding to each camera, the first spatial grid information, the second spatial grid information, the preset reference object, and the key geographic location corresponding to the reference object. In this way, when it is necessary to obtain information related to the geographic location, it can be obtained from the GIS service module.

[0214] In this embodiment, after obtaining the final recognition result, the geographic location information of the target person is obtained according to the image to be recognized, and the preset person trajectory library is updated according to the geographic location information of the target person, so that the preset person trajectory library is kept up to date and the recognition accuracy is improved. In addition, in this embodiment, the geographic location information of the target person is determined through spatial grid information, reference objects, etc., which can improve the recognition accuracy of the geographic location.

[0215] In this embodiment, based on the above-mentioned identity recognition method embodiment, an identity recognition device is provided, which is applied to a monitoring system. The monitoring system includes at least two cameras. Figure 7 , Figure 7 This is a schematic diagram of a module of an identity recognition device of the present invention, and the identity recognition device includes:

[0216] The first acquisition module 71 is used to acquire the image to be recognized collected by the camera at a preset time;

[0217] A first determination module 72, used to determine a human body to be identified from the image to be identified;

[0218] The second acquisition module 73 is used to acquire the human features of the human body to be identified;

[0219] The third acquisition module 74 is used to acquire a person who meets a preset condition from a preset human feature database according to the human features of the person to be identified, so as to obtain a set of candidate persons; the preset human feature database includes the identity of the person and the corresponding human features;

[0220] The judgment module 75 is used to judge whether the human body to be identified matches the candidate person in the candidate person set according to a preset image feature set; the preset image feature set includes: a first historical image feature set collected by the same camera and a second historical image feature set collected by different cameras;

[0221] The second determination module 76 is configured to, if yes, use the matched candidate person as the final recognition result of the human body to be recognized.

[0222] It should be noted that the identity recognition device may optionally include corresponding modules to implement other steps of the above-mentioned identity recognition method.

[0223] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0224] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0225] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0226] The above descriptions are only optional embodiments of the present invention, and are not intended to limit the patent scope of the present invention. All equivalent structural changes made using the contents of the present invention's specification and drawings, or directly / indirectly applied in other related technical fields, are included in the patent protection scope of the present invention.

Claims

1. An identity recognition method, characterized in that: Applied to a monitoring system, the monitoring system includes at least two cameras; the identity recognition method includes the following steps: Obtaining an image to be identified captured by the camera at a preset time; Determining a human body to be identified from the image to be identified; Acquiring the human features of the human body to be identified; According to the human features of the human body to be identified, a person who meets the preset conditions is obtained from a preset human feature database to obtain a set of candidate persons; the preset human feature database includes the identity of the person and the corresponding human features; According to a preset image feature set, determining whether the human body to be identified matches a candidate person in the candidate person set; the preset image feature set includes: a first historical image feature set collected by the same camera and a second historical image feature set collected by a different camera; If so, the matched candidate person is used as the final recognition result of the human body to be recognized; The step of judging whether the human body to be identified matches a candidate person in the candidate person set according to a preset image feature set includes: According to a preset image feature set and a preset person trajectory library, determining whether the person to be identified matches a candidate person in the candidate person set; If yes, after the step of taking the matched candidate person as the final recognition result of the human body to be recognized, the identity recognition method further comprises: According to the image to be identified, obtaining geographic location information of a target person, wherein the target person is a person corresponding to the final identification result; According to the geographic location information of the target person, updating the preset person trajectory library; The at least two cameras include a first camera for collecting the image to be identified and at least one second camera; the step of obtaining the geographic location information of the target person according to the image to be identified includes: Acquire first spatial grid information of the first camera, where the first spatial grid information includes spatial grids obtained by dividing a coverage area of ​​the first camera and geographic locations corresponding to the spatial grids; Determine a first geographical location according to the coordinates of the to-be-identified human body in the to-be-identified image and the first spatial grid information; Determine whether the target person is identified in a reference image captured by the second camera, where the reference image is an image captured by the second camera at the preset time; If yes, determining a second geographical location according to the coordinates of the human body corresponding to the target person in the reference image and second spatial grid information corresponding to the second camera, where the second spatial grid information includes spatial grids obtained by dividing the coverage area of ​​the second camera and geographical locations corresponding to the spatial grids; An overlapping area between the first geographical location and the second geographical location is obtained as the geographical location information of the target person.

2. The identity recognition method according to claim 1, characterized in that: After the step of acquiring the image to be identified captured by the camera at a preset time, the identity recognition method further includes: Evaluating the image quality of the image to be recognized to obtain an image quality evaluation value; Determining whether the image quality assessment value is greater than a preset image quality threshold; If not, preprocessing the image to be identified; The step of determining the human body to be identified from the image to be identified comprises: A human body to be identified is determined from the preprocessed image to be identified.

3. The identity recognition method according to claim 2, characterized in that: The step of preprocessing the image to be identified includes at least one of the following methods: Deblurring the image to be identified; Performing texture enhancement processing on the image to be identified; Performing denoising processing on the image to be identified; Perform light balance processing on the image to be recognized.

4. The identity recognition method according to claim 1, characterized in that: The step of obtaining the human features of the human body to be identified comprises: At least two human features of the human body to be identified are obtained, and the categories of the human features include: facial features, human attribute features, and pedestrian features.

5. The identity recognition method according to claim 1, characterized in that: If yes, after the step of taking the matched candidate person as the final recognition result of the human body to be recognized, the identity recognition method further comprises: The preset image feature set and the preset human feature database are updated according to the human features of the human body to be identified and the final identification result.

6. The identity recognition method according to claim 1, characterized in that: After the step of determining whether the target person is identified in the reference image captured by the second camera, the identity recognition method further includes: If not, the pixel values ​​of the pixels corresponding to the human body to be identified in the image to be identified are set to zero to obtain a background image; Determining whether an object is detected from the background image; If not, taking the first geographical location as the geographical location information of the target person; If so, comparing the object with a preset reference object; If the object does not match the preset reference object, taking the first geographical location as the geographical location information of the target person; If the object matches the preset reference object, determining a third geographical location according to the coordinates of the object in the image to be identified, the coordinates of the human body to be identified in the image to be identified, and the key geographical location corresponding to the preset reference object; An overlapping area between the first geographical location and the third geographical location is obtained as the geographical location information of the target person.

7. The identity recognition method according to claim 1, characterized in that: The step of judging whether the human body to be identified matches a candidate person in the candidate person set according to a preset image feature set and a preset person trajectory library comprises: Screening the candidate personnel in the candidate personnel set according to a preset personnel trajectory library; According to a preset image feature set, it is determined whether the human body to be identified matches the candidate persons in the screened candidate person set.

8. The identity recognition method according to claim 1, characterized in that: The first historical image feature set includes a first image captured by the first camera at the preset moment and a first storage time period before the preset moment, a first human feature extracted from the first image, and a person corresponding to the first human feature; the second historical image feature set includes a second image captured by the second camera at the preset moment and a second storage time period before the preset moment, a second human feature extracted from the second image, and a person corresponding to the second human feature; The step of judging whether the human body to be identified matches a candidate person in the candidate person set according to the preset image feature set includes at least one of the following methods: According to the preset image feature set, determining whether the candidate person exists among the persons corresponding to the previous frame image, the previous frame image being the first image captured before the preset time; According to the preset image feature set, determining whether the candidate person exists among persons corresponding to a first associated image, wherein the first associated image is an image captured by the second camera at the preset time and in the same area as that covered by the first camera; According to the preset image feature set, determining whether the candidate person does not exist among persons corresponding to a second associated image, where the second associated image is an image captured by a second camera at a different coverage area than the first camera at the preset time; According to the preset image feature set, it is determined whether the coverage areas of a target camera and the first camera in the image feature set are adjacent, wherein the target camera is a camera in which the candidate person exists in the second image captured outside the preset time.

9. An identity recognition device, characterized in that: Applied to a monitoring system, the monitoring system includes at least two cameras, and the identity recognition device includes: A first acquisition module, used to acquire the image to be recognized collected by the camera at a preset time; A first determination module is used to determine a human body to be identified from the image to be identified; A second acquisition module is used to acquire the human features of the human body to be identified; A third acquisition module is used to acquire a person who meets a preset condition from a preset human feature database according to the human features of the person to be identified, so as to obtain a set of candidate persons; the preset human feature database includes the identity of the person and the corresponding human features; A judgment module, used to judge whether the human body to be identified matches a candidate person in the candidate person set according to a preset image feature set; the preset image feature set includes: a first historical image feature set collected by the same camera and a second historical image feature set collected by different cameras; A second determination module is used for taking the matched candidate person as the final recognition result of the human body to be recognized; The judgment module is further used to judge whether the human body to be identified matches a candidate person in the candidate person set according to a preset image feature set and a preset person trajectory library; The at least two cameras include a first camera and at least one second camera that capture the image to be identified; the identity recognition device also includes: An updating module is used to obtain geographic location information of a target person according to the image to be identified, and to update the preset person trajectory library according to the geographic location information of the target person, wherein the target person is the person corresponding to the final recognition result; wherein the obtaining the geographic location information of the target person according to the image to be identified includes: obtaining first spatial grid information of the first camera, wherein the first spatial grid information includes spatial grids obtained by dividing the coverage area of ​​the first camera and geographic locations corresponding to each spatial grid; determining a first geographic location according to coordinates of the human body to be identified in the image to be identified and the first spatial grid information; determining whether the target person is identified in a reference image captured by the second camera, wherein the reference image is an image captured by the second camera at the preset time; if so, determining a second geographic location according to coordinates of the human body corresponding to the target person in the reference image and second spatial grid information corresponding to the second camera, wherein the second spatial grid information includes spatial grids obtained by dividing the coverage area of ​​the second camera and geographic locations corresponding to each spatial grid; obtaining an overlapping area of ​​the first geographic location and the second geographic location as the geographic location information of the target person.

10. A terminal device, characterized in that: The terminal device comprises: a memory, a processor, and an identity recognition program stored in the memory and running on the processor, wherein the identity recognition program, when executed by the processor, implements the steps of the identity recognition method according to any one of claims 1 to 8.

11. A storage medium, characterized in that: The storage medium stores an identity recognition program, and when the identity recognition program is executed by the processor, the steps of the identity recognition method according to any one of claims 1 to 8 are implemented.

Images